Plastin 3 rescues cell surface translocation and activation of TrkB in spinal muscular atrophy

Proximal spinal muscular atrophy (SMA) is the most common form of motoneuron diseases in children and young adults, which is caused by the deficiency of the SMN protein (Lefebvre et al., 1997). Although genetic therapies have been successfully applied to SMA patients, it remains enigmatic how SMN loss affects cellular differentiation and maturation of motoneurons and why it can be compensated by SMA modifying proteins such as Plastin 3 (PLS3; Oprea et al., 2008). Primary motoneurons from SMA type I mouse models exhibit reduced β-actin levels that result in decreased growth cone size and impaired axon elongation (Jablonka et al., 2007; Moradi et al., 2017; Nölle et al., 2011; Rossoll et al., 2003; Saal et al., 2014). Furthermore, disturbed accumulation of the voltage-gated calcium channels (VGCCs) Ca_v2.1/2 coincides with reduced spontaneous Ca²⁺ transients (Jablonka et al., 2007) and neurotransmission defects in SMA mice (Tejero et al., 2020; Tejero et al., 2016). Since the orchestration of the presynaptic compartment depends on a well-organized cytoskeleton, decreased Ca_v2.1/2 cluster formations imply impaired localization of further transmembrane proteins in SMA axon terminals.

The functional deficiency of Tropomyosin receptor kinase B (TrkB), the high-affinity receptor for brain-derived neurotrophic factor (BDNF; Chao and Hempstead, 1995; Ebendal, 1992; Reichardt, 2006), causes disturbed differentiation of primary motoneurons (Dombert et al., 2017). Neurotrophic factor signaling activates various signaling pathways indispensable for motoneuron survival and differentiation (Arakawa et al., 1990; Henderson et al., 1994; Hughes et al., 1993; Pennica et al., 1996; Reichardt, 2006; Sendtner et al., 1992a; Sendtner et al., 1992b). The morphological and functional alterations under TrkB kinase domain-deficient conditions such as impaired motoneuron differentiation and reduced spontaneous Ca²⁺ transients phenocopy those observed in SMA motoneurons (Dombert et al., 2017; Jablonka et al., 2007; Rossoll et al., 2003). This inevitably raises the question of whether BDNF/TrkB signaling is altered in SMA. The affected cytoskeleton and unbalanced F- to G-actin ratio in Smn-deficient motoneurons (Jablonka et al., 2007; Moradi et al., 2017; Nölle et al., 2011; Rossoll et al., 2003; Saal et al., 2014) suggest that BDNF/TrkB signaling could be modified through F-actin assembly by modulating proteins such as PLS3. PLS3 associates with the F-actin cytoskeleton within filopodia, where it supports F-actin bundling (Garbett et al., 2020; Giganti et al., 2005) and thus strengthens the protrusive actin network (Garbett et al., 2020). Transcriptome-wide differential expression analysis from SMN1-deleted siblings with discordant disease outcomes revealed a significant association between disease severity and PLS3 expression in female SMA patients (Oprea et al., 2008).

Here, we showed reduced levels and impaired BDNF-induced phosphorylation of TrkB within axon terminals of Smn-deficient motoneurons in vitro and in the neuromuscular junction (NMJ) of a severe SMA mouse model. Virus-mediated human PLS3 (hPLS3) overexpression in cultured SMA motoneurons resulted in a significant rescue of F-actin dynamics that corresponds to improved TrkB translocation and its activation through BDNF. Beyond that, spontaneous Ca²⁺ transients due to improved Ca_v2.1/2 “cluster-like” formations can be restored to wild-type level in vitro. Hence, our study highlights that the affected actin cytoskeleton due to PLS3 downregulation in SMA motoneurons modulates neurotrophic factor signaling and spontaneous Ca²⁺ influx that is crucial for proper cellular differentiation, such as axon elongation and growth cone morphology and functional maintenance of motoneurons.

In order to investigate whether BDNF/TrkB signaling is disturbed upon Smn deficiency and could therefore contribute to SMA pathology, we analyzed NMJs in the affected Transversus abdominis anterior (TVA) muscle of the SMNΔ7 mouse model (Tejero et al., 2016; Torres-Benito et al., 2011) and in growth cones of cultured primary motoneurons of Smn^−/−;SMN2 mouse embryos. For this purpose, NMJs were analyzed at an early symptomatic stage at postnatal day (P) 5 and a later symptomatic stage at P10-11. Consistent with previous studies (Hosseinibarkooie et al., 2016; Ruiz et al., 2010; Tejero et al., 2020; Tejero et al., 2016), the total presynaptic area as well as the ratio of presynaptic and postsynaptic areas was significantly reduced in SMNΔ7 pups at P10-11 (Fig. S1 A). To determine TrkB localization and activation at axon terminals, we performed immunostainings with TrkB and phosphorylated TrkB (p-TrkB) antibodies. Specificity was verified by analysis of Ntrk2^−/− NMJs at P5 (Fig. S2 A) and growth cones of cultured motoneurons from Ntrk2^−/− mouse embryos (Fig. S2, B–D). In SMNΔ7 presynaptic terminals, the total TrkB levels were markedly decreased already at P5 (Fig. S1 B), and this became even more pronounced (∼50%) during disease progression at P10-11 (Fig. 1 A). Interestingly, SMA motoneurons showed only a moderate but significant loss in total TrkB levels in growth cones (Fig. 1 B). In whole-cell extracts, transcript quantification through RT-PCR (Fig. 1 C) and Western blot analysis (Fig. S1 C) did not show any differences in TrkB levels between Smn-deficient and control motoneurons. These results suggest that alterations in TrkB localization exclusively relate to cellular dysregulations in axon terminals of SMA motoneurons.

To examine the activation of the receptor upon stimulation with its ligand, p-TrkB immunostainings were performed on TVA muscle explants that were BDNF-stimulated for 30 min prior to fixation. At P10-11, significantly reduced levels of p-TrkB have been observed, and BDNF stimulation could not evoke a raise in p-TrkB signals in SMNΔ7 mice (Fig. 1 D). Although significantly decreased levels of p-TrkB have been observed already at P5, BDNF treatment could still evoke a marked increase of p-TrkB in the endplates of early symptomatic SMNΔ7 mice (Fig. S1 D). In contrast to Western blot analysis of whole-cell extracts (Fig. S1 C), immunocytochemical analysis of Smn-deficient growth cones revealed an impaired BDNF-induced TrkB phosphorylation in comparison with control growth cones (Fig. 1 E). To examine whether the alterations of the actin cytoskeleton in SMA axon terminals are involved in the affected TrkB phosphorylation, motoneurons were treated with the mycotoxin Cytochalasin D (CytoD) prior to BDNF stimulation to inhibit actin polymerization. The analysis of p-TrkB in axon terminals revealed that the BDNF-induced TrkB activation is absent upon CytoD pre-treatment (Fig. 1 F). Hence, the disruption of the cytoskeleton phenocopies Smn deficiency and seems to be causative for the impaired TrkB activation. Collectively, these data argue that loss of Smn interferes with localization and activation of TrkB in the axon terminals and motor endplates that consequently could lead to failures in the BDNF/TrkB-signaling cascade.

To examine the possibility that the impaired TrkB activation is due to altered receptor presentation at the cell surface, a live-staining protocol was established to label TrkB exclusively localized at the cell surface (see Materials and methods). No differences were observed in baseline TrkB surface levels in Smn-deficient motoneurons compared to controls (Fig. 2 A, first lane). Next, we studied whether endocytosis of the receptor upon BDNF stimulation is disturbed in Smn-deficient axon terminals. Therefore, cultured motoneurons were stimulated with BDNF for 5 min and 15 min. TrkB disappears from the cell surface following BDNF stimulation in a time-dependent manner in control motoneurons, and similar observations were made for SMA growth cones (Fig. 2 A). Thus, BDNF-mediated TrkB internalization is not altered in SMA motoneurons. Early studies with motoneurons have shown that cyclic AMP (cAMP) is one of the key regulators of TrkB surface presentation and therefore modulates the responsiveness to BDNF (Du et al., 2000; Meyer-Franke et al., 1998). To investigate this mechanism in SMA motoneurons, we used cAMP to stimulate TrkB transfer to the plasma membrane. Indeed, a 20 min 8-CPT-cAMP stimulation was sufficient to induce a robust elevation of TrkB at the cell surface of control motoneurons. However, no shift in TrkB translocation to the plasma membrane was observed in Smn-deficient growth cones (Fig. 2 B). The failure of 8-CPT-cAMP to recruit TrkB to the cell surface in SMA motoneurons could be caused by a defective actin cytoskeleton. To address this possibility, motoneurons were pre-treated with CytoD prior to 8-CPT-cAMP stimulation that prevented the cAMP-induced TrkB surface translocation in control motoneurons, while no further effect could be observed in Smn-deficient cells (Fig. 2 C). Thus, our experiments demonstrate that the ligand-induced TrkB internalization in SMA motoneurons is unaffected but the cAMP-mediated TrkB recruitment to the cell surface through actin filaments is impaired.

In order to explore compartment-specific transcript alterations that relate to the dysregulated actin cytoskeleton and therefore could be relevant for defective TrkB translocation and activation, we performed RNA sequencing (RNA-seq) of Smn knockdown (shSmn) motoneurons (Fig. 3 A) cultured in compartmentalized microfluidic chambers. The differential gene expression analysis (Table S1) revealed a variety of transcripts that showed altered expression upon Smn knockdown in the somatodendritic and the axonal compartment compared to the GFP-expressing controls (Fig. 3 B). 604 transcripts showed a significant (P < 0.05) down- or upregulation in the somatodendritic compartment, whereas 521 transcripts were significantly down- or upregulated in the axonal compartment of Smn-deficient motoneurons (Fig. 3 C), with an intersection of 132 transcripts from both compartments (Fig. 3, C and D; and Table S1). We further used Gene Ontology (GO)-term analysis to examine pathways potentially altered in the two compartments (Fig. 3 E and Table S2) of Smn knockdown motoneurons. Interestingly, ATP- and oxidative phosphorylation-related genes are upregulated in both compartments, while especially GTP-linked and microtubule-associated candidate transcripts are downregulated in the somatodendritic compartment (Fig. 3 E and Table S2). Only in the axonal compartment, transcripts encoding proteins with functions related to the actin cytoskeleton appear to be downregulated in Smn-deficient motoneurons (Fig. 3, E–G). Within the top 10 of axonal downregulated actin-related transcripts (Fig. 3, F and G, yellow underlay) we identified the actin-modulator Pls3 (Fig. 3, F–H). Different levels of PLS3 are a characteristic feature in SMN1-deleted siblings with discordant disease severity (Oprea et al., 2008). Pls3 is a potent actin-bundler required for F-actin stability (Karpova et al., 1995; Shinomiya, 2012). The importance of a stabilized actin cytoskeleton for correct trafficking of the presynaptic ER or vesicles derived from the ER was already shown (Deng et al., 2021). Thus, the reduced levels of Pls3 might interfere with correct translocation of transmembrane proteins such as TrkB. In addition to this, Arpc1b, a component of the actin-related 2/3 (Arp2/3) complex, which is important for the initiation of actin filament branching (Lee et al., 2010; Suraneni et al., 2012; Welch et al., 1997; Wu et al., 2012a), is also found among the 10 most downregulated actin-related transcripts in the axonal compartment (Fig. 3, F and G). These data indicate that Smn deficiency impacts actin-related transcripts in the axonal compartment.

As RNA-seq data revealed that the actin-bundling protein Pls3, the most relevant genetic SMA modifier, is markedly decreased in the axonal compartment of Smn knockdown motoneurons, we asked whether a lack of Pls3 could be involved in impaired TrkB localization and activation in axon terminals of Smn-deficient motoneurons. Quantification of the Pls3 immunosignal within growth cones confirmed the RNA-seq data of Smn knockdown motoneurons, since Smn^−/−;SMN2 growth cones showed a significant reduction of Pls3 (Fig. 4 A). To test whether Pls3 is involved in actin-related processes modulating the translocation and activation of TrkB, we used a shRNA approach to target the endogenous Pls3 (shPls3) and used Luciferase-shRNA (shLuci) as control. Quantitative transcript analysis through RT-PCR confirmed a knockdown efficiency of ∼60% in motoneurons transduced with the shPls3-harboring lentivirus (Fig. 4 B). Morphological analysis revealed that, like SMA motoneurons (Jablonka et al., 2007), the axons of Pls3 knockdown cells cultured on laminin-221/211 exhibited altered elongation (Fig. 4 C) as well as reduced growth cone size (Fig. 4 D). Since PLS3 acts as an F-actin-bundling protein, we used live-cell imaging to assess how the reduced Pls3 levels modulate actin dynamics. For this purpose, cultured Pls3 knockdown motoneurons were incubated with a fluorescent F-actin binding probe (SiR-actin) to visualize actin movements over a period of 20 min (Fig. 4, E–G). Quantification of the SiR-actin signal revealed that Pls3 knockdown motoneurons display impaired actin movements in the growth cone, as shown by reduced moving distance of single filopodia (Fig. 4 E) and decreased velocity of the movements (Fig. 4 F). To examine BDNF/TrkB signaling in these neurons, we investigated the TrkB localization in Pls3-depleted growth cones. Interestingly, we found significantly reduced total TrkB levels (Fig. 4 H). Comparison of the ligand-induced immunosignal of p-TrkB after 15 min BDNF stimulation revealed that, comparable to Smn-deficient neurons, the activation of the receptor is impaired (Fig. 4 I). Next, we examined the responsiveness to BDNF and cAMP stimulation of receptors presented at the cell surface. In our live-staining assay, we observed preserved baseline surface TrkB levels and an intact TrkB endocytosis upon 15 min BDNF stimulation (Fig. 4 J). However, we did not monitor any 8-CPT-cAMP-induced TrkB surface recruitment in Pls3-deficient motoneurons (Fig. 4 J). In conclusion, these experiments imply that Pls3 is functionally participating in the dynamic surface presentation of TrkB and its activation through BDNF stimulation.

Based on these results, we then hypothesized that the membrane translocation and accumulation of VGCCs at presynaptic terminals might as well be actin dependent. Thus, we investigated whether Pls3 knockdown interferes with the localization and function of Ca_v2.2 in axon terminals. Quantification of the Ca_v2.2 immunosignal in Pls3-deficient growth cones showed significantly reduced levels of Ca_v2.2 (Fig. 4 K). Moreover, high resolution Structured Illumination Microscopy (SIM) revealed that the formation of Ca_v2.2 “cluster-like” accumulations are markedly diminished upon knockdown of Pls3 (Fig. 4 L, white arrows), similar to Smn-deficient motoneurons. Hence, we asked whether the functionality of the VGCCs is impaired in these neurons. Therefore, we examined the frequency of spontaneous spike-like Ca²⁺ transients in the growth cones of Pls3-depleted motoneurons. Indeed, we found that in contrast to the GFP-expressing control motoneurons, Pls3 knockdown growth cones display reduced spontaneous frequencies of Ca²⁺ transients in their axon terminals (Fig. 4 M). Thus, our data demonstrate that Pls3 is required for the correct alignment and function of other presynaptic transmembrane proteins, such as VGCCs, in addition to TrkB.

Since Pls3 knockdown and SMA motoneurons show similar defects in axonal elongation, TrkB phosphorylation, cAMP-mediated TrkB surface translocation, and cellular excitability, we studied the effects of hPLS3 overexpression in SMA axon terminals. Smn-deficient motoneurons were transduced with the human PLS3-expressing lentivirus (LV-hPLS3) or a control virus delivering mCherry (LV-mCh). Quantitative analysis revealed a robust overexpression of hPLS3 mRNA (Fig. 5 A). Intriguingly, overexpressing hPLS3 in SMA motoneurons restored impaired axon elongation (Fig. 5 B) and growth cone size (Fig. 5 C) to wild-type levels. In control motoneurons, the overexpression of hPLS3 had no effects on axon growth and differentiation. To study whether elevated hPLS3 levels modulate actin dynamics, we visualized actin movements over a period of 20 min using SiR-actin (Fig. 5, D–F). Quantification of the SiR-actin signal revealed that SMA motoneurons display reduced actin movements in the axon terminals (Fig. 5, D and E). The distance of individual filopodia movements (Fig. 5 D), as well as the velocity of these movements (Fig. 5 E), were significantly enhanced upon hPLS3 overexpression. Analysis of the ratio of F/G-actin revealed that Smn-deficient cells transduced with LV-mCh display diminished levels of F-actin, which is reflected by a reduced F/G-actin ratio (Fig. 5 G). Again, overexpression of hPLS3 improved this effect (Fig. 5 G). These data indicate that overexpressing hPLS3 in SMA motoneurons causes marked morphological and functional improvements. We then tested whether the observed TrkB defects are ameliorated as well. Quantification of the TrkB immunofluorescence signal in Smn-deficient cells using confocal microscopy (Fig. 5 H) and SIM (Fig. 5 I) revealed that total TrkB levels are slightly increased upon hPLS3 overexpression. Moreover, we used SIM to study the localization of Pls3 and TrkB along actin filaments in axon terminals. In control motoneurons (LV-mCh and LV-hPLS3 transduced), Pls3 was located along bundled actin filaments (Fig. 5, J and K). However, SIM confirmed previous findings showing reduced levels of F-actin, Pls3, and TrkB in LV-mCh transduced SMA motoneurons (Fig. 5 L). Upon overexpression of hPLS3, the morphology of Smn-deficient growth cones reversed back to that resembling control motoneurons. The level of bundled actin filaments is enhanced and showed abundant localization of Pls3 along the filaments (Fig. 5 M). Interestingly, TrkB is mostly located in proximity of Pls3-bundled actin filaments, pointing to a role of these filaments in TrkB surface translocation. Therefore, we checked TrkB expression on the cell surface. Comparison of the baseline surface TrkB levels revealed no differences; however, when hPLS3-overexpressing Smn-deficient motoneurons were treated with 8-CPT-cAMP to induce receptor recruitment to the cell surface, a marked rise of TrkB at the plasma membrane was observed, while no such increase was detectable in LV-mCh transduced SMA motoneurons (Fig. 6 A). These findings indicate that the impaired cAMP-induced TrkB membrane translocation can be rescued by hPLS3 overexpression in SMA motoneurons. Further evidence for improved dynamical vesicular TrkB transport upon cAMP stimulation along actin filaments in Smn-deficient motor axons overexpressing hPLS3 is qualitatively documented by Videos 1, 2, and 3. Since hPLS3 overexpression normalizes the distribution of TrkB and its activity-dependent surface translocation, we assessed whether the BDNF-induced TrkB activation is rescued as well. Comparison of the p-TrkB immunosignal after 15 min BDNF stimulation showed that LV-mCh transduced Smn-deficient motoneurons did not exhibit increased p-TrkB immunoreactivity. This defect was normalized to control levels when hPLS3 was overexpressed (Fig. 6 B). To test whether TrkB activation leads to functional downstream signal transduction, we investigated the phosphorylation of Akt (p-Akt) upon BDNF stimulation. We could demonstrate that a 15 min BDNF stimulation is sufficient to increase the p-Akt signal in control cells, independent of viral transduction, but was significantly less effective in Smn-deficient motoneurons (Fig. 6 C). However, hPLS3-overexpressing SMA motoneurons showed a much stronger increase of p-Akt upon BDNF stimulation (Fig. 6 C). Taken together, hPLS3 overexpression in SMA motoneurons rescues the localization and cAMP-induced translocation as well as the activation of TrkB leading to enhanced BDNF/TrkB signaling.

Based on our previous results on TrkB and the fact that Pls3 knockdown interferes with Ca_v2.2 localization and function, we examined whether hPLS3 overexpression can rescue disturbed Ca_v2.2 “cluster-like” formations in SMA motoneurons (Jablonka et al., 2007). As shown before (Jablonka et al., 2007), Ca_v2.2 immunoreactivity was significantly reduced in LV-mCh transduced Smn-deficient neurons (Fig. 7 A). Intriguingly, hPLS3 overexpression improved the levels of Ca_v2.2 in these terminals (Fig. 7 A). Furthermore, we used SIM to test whether the “cluster-like” accumulations of these VGCCs is restored as well. Indeed, comparison of “cluster-like” Ca_v2.2 structures revealed less of these accumulations in SMA motoneurons transduced with LV-mCh, while multiple of these structures could be observed in hPLS3-overexpressing SMA motoneurons (Fig. 7 B). Next, we examined the functionality of the VGCCs using the calcium imaging technique. LV-mCh transduced Smn-deficient motoneurons showed significantly reduced frequencies of spontaneous spike-like Ca²⁺ transients (Fig. 7 C) in comparison with control motoneurons. In contrast, overexpression of hPLS3 markedly increased the frequency of spike-like spontaneous Ca²⁺ transients in the axon terminals of SMA motoneurons to normal levels observed in control motoneurons (Fig. 7 C). Hence, hPLS3 overexpression in SMA motoneurons improves “cluster-like” formations of Ca_v2.2 and enhances frequency of spontaneous Ca²⁺ transients.

In SMA motoneurons, basal levels of surface TrkB and endocytosis after BDNF stimulation seems to be unchanged. However, the cAMP-mediated membrane translocation is highly impaired upon Smn deficiency. Therefore, the question arises why the raise of p-TrkB after a 15 min BDNF stimulation is abolished in Smn-deficient axon terminals compared to controls. A possible mechanism that could explain the failure in TrkB activation is based on the self-amplifying actions of BDNF, since it was shown to be capable of triggering TrkB surface insertion for signal amplification (Cheng et al., 2011). Thus, BDNF stimulation leads to activation and internalization of TrkB, but could also promote in a positive feedback loop further recruitment of TrkB to the plasma membrane (Andreska et al., 2020; Cheng et al., 2011). Based on our observations that CytoD treatment abolished cAMP-mediated TrkB translocation to the cell surface (Fig. 2 C), the TrkB recruitment is dependent on a proper working actin cytoskeleton. Thus, the self-amplifying mechanism by local enhancement of available TrkB that is activated by BDNF stimulation seems to be impaired in SMA motoneurons, which would explain the decreased p-TrkB levels. To test this hypothesis, we investigated re-constitution of cell surface TrkB levels after BDNF stimulation. Control and Smn-deficient motoneurons were stimulated with BDNF for 5 min, then BDNF-containing medium was washed out, and the neurons were allowed to recover for 10 min. The comparison of TrkB surface levels showed that upon 5 min BDNF stimulation, the receptors disappeared from the cell surface (Fig. 8 A) as shown before (Fig. 2 A). However, following a 10 min recovery phase, the levels of TrkB at the cell surface started to reconstitute in axon terminals of control cells (Fig. 8 A). Thus, this assay reveals that more TrkB is re-located to the cell surface after BDNF-induced endocytosis of the receptor. In SMA growth cones, the recruitment of TrkB to the cell surface during the recovery phase failed (Fig. 8 A). To confirm that this process is dependent on a functional actin cytoskeleton, control motoneurons were treated with CytoD, Nocodazole (which blocks microtubule polymerization), or Cycloheximide (which interferes with protein synthesis) prior to the 10 min recovery assay. CytoD pre-treatment prevented the reconstitution of TrkB to the cell surface during the 10 min recovery phase, while disrupting the microtubules with Nocodazole had only minor effects (Fig. 8 B). Moreover, these data show that the receptors are recruited from an already existing intracellular pool rather than being newly synthetized, since protein translational elongation blockage by Cycloheximide had no effect on the recovery (Fig. 8 B). This was further confirmed by a Pls3 knockdown in wild-type motoneurons. Depletion of Pls3 prevented the reconstitution of surface TrkB levels during the recovery phase (Fig. 8 C) in a similar manner as in Smn-deficient motoneurons. In line with this, hPLS3 overexpression in SMA motoneurons showed a strong rescue of the defects in TrkB translocation to the cell surface (Fig. 8 D). Next, we asked whether the rapid recovery of surface TrkB is based on recycled receptors that had been on the surface prior to BDNF stimulation, or whether they originate from the intracellular pool of TrkB. For this purpose, we performed a TrkB recycling assay on living motoneurons (Fig. 8 E; see Materials and methods). Motoneurons were exposed to a TrkB antibody and secondary antibody prior to BDNF stimulation, and a second TrkB staining was performed after the recovery phase to directly follow TrkB fate within individual growth cones (Fig. 8 E). While in control motoneurons the baseline levels of TrkB are reconstituted up to ∼67%, only ∼40% of TrkB are detectable at the cell surface upon recovery in Smn-deficient motoneurons. Improving the actin dynamics by hPLS3 overexpression rescued the recovery in SMA axon terminals (∼72%) to control level (Fig. 8 F). Interestingly, we observed that approximately half of the recovered TrkB have previously been at the surface in control motoneurons (57.7%) as well as in Smn-deficient cells (55.8%). Thus, this assay shows that the highly impaired TrkB recovery in SMA motoneurons is due to an actin-dependent translocation defect to the cell surface rather than disturbed surface TrkB recycling per se.

In conclusion, our data support the hypothesis that TrkB is rapidly translocated to the plasma membrane after ligand-induced internalization and that this process depends on an intact actin cytoskeleton.

To gain further insights into the role of PLS3 in promoting F-actin-mediated surface translocation of transmembrane proteins, we examined the interaction with additional actin-modulating factors. One candidate is the actin-related protein 2/3 (Arp2/3) complex, which was recently shown to be markedly diminished at the leading edge of endothelial cells upon loss of PLS3 (Garbett et al., 2020). Furthermore, transcripts encoding for the complex components Arpc1b and Arpc4 appeared to be downregulated in the axonal compartment of Smn knockdown motoneurons (Fig. 3 G). Arp2/3 is an actin-nucleator that initiates the formation of branched actin filaments (Lee et al., 2010; Suraneni et al., 2012; Welch et al., 1997; Wu et al., 2012a), which are primarily bundled by PLS3 (Skau et al., 2011). We first checked the levels of the Arp2/3 complex in the growth cones of control and Smn-deficient motoneurons with an Arp3 antibody. Interestingly, the expression of Arp3 was reduced within axon terminals of SMA motoneurons (Fig. 9 A). SIM images revealed that Pls3 is distributed all over the growth cone but is highly enriched within F-actin-rich protrusions, while Arp3 is mainly located in the core of the growth cone (Fig. 9 B). The expression of Arp3 was also significantly reduced in growth cones upon knockdown of Pls3 in wild-type motoneurons (Fig. 9 C). In turn, overexpression of hPLS3 in Smn-deficient motoneurons resulted in restoration of the Arp3 signal to control level (Fig. 9 D). To better understand the relationship of Arp2/3-mediated actin-branching and TrkB membrane translocation, we pre-treated motoneurons with the Arp2/3 inhibitor CK-666 and used our live-cell staining protocol. Intriguingly, addition of CK-666 before and during BDNF stimulation and recovery phase prevented the recruitment of TrkB in control motoneurons, while no further effect was observed in SMA motoneurons (Fig. 9 E). Next, we wanted to figure out whether the beneficial effects of hPLS3 overexpression on translocation of TrkB upon BDNF stimulation are abolished by Arp2/3 inhibition. Indeed, we found that CK-666 treatment abolished the rescuing effect of hPLS3 overexpression in Smn-deficient motoneurons. No TrkB recruitment to the cell surface was detectable anymore upon recovery (Fig. 9 F). Altogether, these results indicate that a well-balanced cooperation of the Arp2/3 complex and Pls3 is required for the function of the actin cytoskeleton in mediating the ligand-induced translocation of TrkB to the cell surface. In addition to the members of the Arp2/3 complex Arpc4 and Arpc1b, RNA-seq analysis revealed the downregulation of Pfn1 transcripts in the axonal compartment of Smn knockdown motoneurons (Fig. 3 F). Profilin inherits complex functions and is involved in the regulation of directed actin polymerization (Hensel and Claus, 2018; Witke, 2004). Moreover, is has been shown that altered profilin protein levels could be involved in the pathogenesis of SMA (Antoine et al., 2020; Bowerman et al., 2007). Hence, we examined whether overexpression of hPLS3 beneficially acts on the regulation of profilin. Therefore, we quantified levels of profilin using a pan-profilin antibody in motor axon terminals. SMA motoneurons transduced with LV-mCh displayed significantly decreased levels of profilin immunoreactivity, which were significantly enhanced upon overexpression of hPLS3 (Fig. S3 A). Since the activity of profilin is regulated by phosphorylation, we also tested the phosphorylation status of profilin (p-profilin). Under standard conditions, no alterations in p-profilin levels were observed in SMA axon terminals (Fig. S3 B). Recently, profilin phosphorylation was shown to be stimulated by BDNF (Dombert et al., 2017). Therefore, motoneurons were exposed to BDNF for 15 min, and the p-profilin levels were examined. In control motoneurons transduced with LV-mCh or LV-hPLS3, a marked increase in p-profilin quantities was detected upon BDNF stimulation (Fig. S3 B). In contrast to that, no such increase was measurable in LV-mCh transduced Smn-deficient motoneurons. Again, when hPLS3 was overexpressed, the BDNF-induced profilin phosphorylation was normalized to control levels (Fig. S3 B). In contrast to that, the levels of cofilin (Fig. S3 C), an actin-depolarization factor (Sarmiere and Bamburg, 2004), or myosin VI, which acts in a complex with TrkB for the retrograde signaling (Yano et al., 2006a), were not altered in SMA motoneurons (Fig. S3 D).

In order to investigate the effect of hPLS3 on stabilizing the actin cytoskeleton in motor nerve terminals in vivo, we overexpressed hPLS3 through adeno-associated viruses 9 (AAV9). SMNΔ7 mice were injected with AAV9-hPLS3 or AAV9-EGFP at P0/P1, and the NMJs of the animals were analyzed at the later-symptomatic stage P10-P11. Overexpression of hPLS3 was verified in motoneurons through HA-tag within the spinal lumbar L1 segment (Fig. S4). Morphological investigations revealed an increased ratio of occupancy of postsynaptic areas by presynaptic terminals upon hPLS3 overexpression in Smn-deficient NMJs (Fig. 10 A). Examination of the amount of G-actin within the presynaptic terminals showed significantly lower levels in Smn-deficient endplates, while overexpression of hPLS3 elevated the level of G-actin (Fig. 10 B). In addition, total TrkB levels within the presynapse increased upon overexpression of hPLS3 in axon terminals of SMNΔ7 mice (Fig. 10 C). To test whether overexpression of hPLS3 manifests in considerable improvements of TrkB signaling at the NMJ, TVA muscle explants were stimulated with BDNF, and the activation of TrkB was measured in the axon terminals. Strikingly, and in contrast to the AAV9-EGFP-injected SMNΔ7 mice, AAV9-hPLS3-treated SMA pups showed a marked increase in p-TrkB immunosignal in the NMJs upon BDNF stimulation (Fig. 10 D). Thus, hPLS3 overexpression exhibits a favorable effect on TrkB localization and ligand-induced activation in Smn-deficient NMJs.

Moreover, Ca_v2.1 “cluster-like” formations were decreased in the neuromuscular endplate of AAV9-EGFP-injected SMNΔ7 mice, which was significantly improved when hPLS3 was overexpressed in these animals (Fig. 10 E). Hence, our in vivo data demonstrate that postnatal overexpression of hPLS3 in a severe SMA mouse model has beneficial effects on the activation of TrkB and partially improves the progressive NMJ deterioration.

Given the tremendous impact of BDNF/TrkB signaling on neuronal survival and differentiation (Kowiański et al., 2018), it is not surprising that dysregulations of this mechanism play a crucial role in various diseases such as Alzheimer’s disease (Amidfar et al., 2020), Parkinson’s disease (Palasz et al., 2020), or epilepsy (Lin et al., 2020). However, the underlying mechanisms vary strongly between different diseases, reaching from neuroprotective to neurotoxic effects, as recently reported for amyotrophic lateral sclerosis (ALS; Pradhan et al., 2019). This makes it challenging to develop approaches for alternative therapeutic strategies based on BDNF/TrkB signaling. Focusing on SMA, we aimed to decipher the cellular mechanisms that underlie altered TrkB cell surface exposure and impaired BDNF-induced activation and signaling. Our study provides evidence that TrkB phosphorylation and downstream signaling such as Akt phosphorylation after BDNF stimulation is reduced in Smn-deficient growth cones, although TrkB presentation at the cell surface under basal conditions is unaltered. In SMA axon terminals, BDNF-mediated TrkB endocytosis is working properly in contrast to the activity-mediated TrkB recruitment to the plasma membrane and its re-translocation upon BDNF-mediated activation. The fact that the levels of myosin VI, which was shown to be responsible for the retrograde transport of BDNF/TrkB signaling endosomes (Yano et al., 2006a), were unaltered in SMA motoneurons further indicates that BDNF/TrkB complex endocytosis is functioning in SMA axon terminals. However, TrkB recycling experiments revealed that baseline TrkB levels are reconstituted after BDNF stimulation up to ∼67% in control cells, while in Smn-deficient motoneuron terminals only ∼40% of baseline TrkB level are observed upon recovery. We concluded that a self-amplifying mechanism of BDNF by locally enhancing the amount of available TrkB at the cell surface for signal amplification corresponds to proper F-actin assembly that is dysfunctional in SMA motor axon terminals. Based on this assumption, we started focusing on PLS3 and its connection to the F-actin cytoskeleton. Although PLS3 has been already shown to beneficially modify the severity of the SMA phenotype in patients and relevant animal models (Ackermann et al., 2013; Alrafiah et al., 2018; Dimitriadi et al., 2010; Hao et al., 2012; Hosseinibarkooie et al., 2016; Lyon et al., 2013; Oprea et al., 2008), the knowledge of the cellular mechanism behind the PLS3-compensatory effect is still fragmented, as recently reviewed (Wolff et al., 2021).

In our study, we could show that Pls3 mRNA is downregulated together with other actin-related transcripts in the axonal compartment of Smn-deficient motoneurons. This underlines our hypothesis that reduced axon elongation of SMA motoneurons involves defects in the F-actin cytoskeleton (Jablonka et al., 2007; Rossoll et al., 2003; Rossoll et al., 2002). In the somatodendritic compartment, we observed a downregulation of transcripts associated with the microtubule cytoskeleton or GTPase activity that is indispensable for axonal transport. This could explain the disturbed membrane coverage of the rectifier channel K_v2.1 (Fletcher et al., 2017) that corresponds to impaired proprioceptive input on SMA somata (Fletcher et al., 2017; Mentis et al., 2011). The upregulation of transcripts whose corresponding proteins are involved in oxidative phosphorylation and ATP metabolism in both compartments of Smn-deficient motoneurons could represent a compensatory mechanism counteracting cytoskeletal defects. Beyond that, neuronal differentiation requires the switch from glycolysis to oxidative phosphorylation for enhanced ATP production (Iwata and Vanderhaeghen, 2021). Since SMA motoneurons show obvious mitochondrial dysregulations due to Ca²⁺ signaling alterations (Lopez-Manzaneda et al., 2021), the upregulation of those transcripts might compensate the mitochondrial dysfunctions and try to correct cellular differentiation.

The yet unanswered question about the cause of defective Ca_v2.2 cluster formations and reduced spontaneous Ca²⁺ transients in SMA motoneurons (Jablonka et al., 2007) seems to have found an answer. Pls3 is a crucial player in this scenario, as the Pls3 knockdown phenocopies the morphological and functional alterations of SMA motoneurons on the β2-chain laminin isoform (laminin-221/211). Beyond that, Pls3 knockdown also affects TrkB localization and phosphorylation that is newly reported in this study for SMA motoneurons. In addition, the importance of PLS3 is already reflected by its role as a genetic SMA modifier and could be confirmed by our rescue experiments with hPLS3 overexpression in Smn-deficient motoneurons. Overexpression of hPLS3 modulates F-actin dynamics and leads to improved cAMP-dependent recruitment of TrkB to the cell surface, as well as to reconstituted spontaneous Ca²⁺ transients based on improved Ca_v2.2 “cluster-like” formations in SMA axon terminals. Strikingly, this effect is comparable with previous observations on the beneficial effects of R-Roscovitine on SMA motoneurons. Application of this drug increases spontaneous Ca²⁺ transients by extending the open state of Ca_v2.2, leading to reconstituted cellular excitability and differentiation (Tejero et al., 2020). Similarly, chronic in vivo treatment of the K⁺ channel blocker 4-aminopyridine, which increases neuronal activity and Ca²⁺ influx at presynaptic terminals, restores NMJ pathology through axonal sprouting in SMA mice (Simon et al., 2021).

Beyond this, the actin-bundling capacity of PLS3 is linked to the Arp2/3 actin nucleator complex (Giganti et al., 2005). Smn-deficient motoneurons are Arp3-reduced at their nerve terminals that significantly improved upon hPLS3 overexpression. More recently, Garbett and colleagues showed that PLS3 knockdown results in a defective protrusive network, including decreased levels of Arp3 and F-actin in lamellipodia (Garbett et al., 2020). In particular, the interplay between Pls3 and the Arp2/3 complex seems to be important for proper function of the actin cytoskeleton in motoneurons, as blockage of the Arp2/3 complex by CK-666 inhibited the rescuing effects of hPLS3 overexpression on TrkB translocation upon recovery. Besides Pls3 and Arp2/3, further regulatory proteins such as profilin or cofilin are required for proper functioning of the actin cytoskeleton. Profilin has specific functions in regulating the inclusion of G-actin to the barbed ends of F-actin (Hensel and Claus, 2018). Thus, it is not surprising that dysregulation of profilin is connected to the pathomechanism of motoneuron diseases such as ALS (Smith et al., 2015; Tanaka et al., 2016; Wu et al., 2012b) and SMA (Antoine et al., 2020; Bowerman et al., 2007; Giesemann et al., 1999). Moreover, in the context of SMA, a link between profilin levels and Pls3 has been already reported (Bowerman et al., 2009). In line with this, RNA-seq and immunocytochemistry revealed a reduction of profilin in SMA axon terminals. The baseline phosphorylation status of profilin was unaltered in SMA growth cones. Based on the defective BDNF/TrkB signaling in SMA motoneurons, we tested whether BDNF stimulation can induce an increase in p-profilin levels as shown previously (Dombert et al., 2017). SMA neurons did not show increased p-profilin levels upon BDNF stimulation. Intriguingly, hPLS3 overexpression resulted in enhanced levels of overall profilin and increased p-profilin levels upon BDNF stimulation in SMA motoneurons. Thus, increasing the levels of Pls3 beneficially influences the localization and regulation of actin regulatory proteins such as Arp2/3 or profilin, which are a prerequisite for proper functioning of the actin cytoskeleton, in particular for the translocation and alignment of transmembrane proteins. However, it is possible that other axonal downregulated transcripts, whose corresponding proteins function in actin-bundling and cross-linking such α-actinin (Actn1) or filamin A (Flna; Stevenson et al., 2012) contribute to the disturbed actin dynamics thereby influencing translocation of transmembrane proteins.

Interestingly, hPLS3 overexpression in vivo resulted in markedly enhanced BDNF-induced TrkB activation and significantly improved Ca_v2.1 “cluster-like” formations in SMN∆7 neuromuscular endplates. However, only minor improvements of TrkB localization, G-actin levels and area ratio of the pre- vs. postsynapse were found in NMJs of these animals upon hPLS3 overexpression. This could be due to the postnatal overexpression of hPLS3, as correct F-actin assembly is required already during early embryonic development when motoneurons differentiate and grow out to their targets to form NMJs. Furthermore, as the SMN∆7 mouse model is a severely affected disease model, the observed morphological and functional defects due to Smn deficiency are probably only moderately restorable through a partial AAV9-hPLS3 overexpression (Ackermann et al., 2013; Alrafiah et al., 2018; Kong et al., 2021; McGovern et al., 2015). This has already been reported for SMA patients, since the protective effect of PLS3 is only observed in SMA type 2 and 3 but not in severely affected children with only one or two SMN2 copies (Oprea et al., 2008) and less severely affected mouse models (Hosseinibarkooie et al., 2016).

In conclusion, our findings reveal that Smn deficiency is associated with the defective activity-induced recruitment of the high-affinity BDNF receptor TrkB to the cell surface at motor axon terminals. As BDNF/TrkB signaling is indispensable for proper motoneuron differentiation and maturation, our data strongly suggest that some of the already described cellular dysregulations are likely to be due to affected BDNF/TrkB signaling. Strikingly, the SMA modifier protein PLS3 is able to rescue TrkB recruitment to the cell surface and to correct localization of VGCCs leading to increased frequency of spontaneous Ca²⁺ transients. Based on our results and encouraged by other studies, we postulate a novel role for PLS3 in supporting correct alignment of transmembrane proteins such as TrkB and Ca_v2.1/2 through modulation of F-actin dynamics in axon terminals of motoneurons. This mechanism is a basic prerequisite for proper motoneuron differentiation and maturation probably also in other neurodegenerative disorders beyond SMA.

Laboratory mice were housed under controlled conditions in the animal facilities of the Institute of Clinical Neurobiology at the University Hospital Würzburg and Faculty of Medicine, University of Leipzig with access to food and water ad libitum. All described procedures were performed by trained personal following the regulations on animal protection of the German federal law and of the Association for Assessment and Accreditation of Laboratory Animal care, as well as the state authorities. For our studies, various wild-type, SMA mouse models and TrkB-deficient mice were used. CD-1 and FVB mice were obtained from Charles River. Smn^+/−;SMN2 (RRID:IMSR_JAX:005024) and Smn^+/−;SMN2;SMNΔ7 mice (RRID:IMSR_JAX:005025), initially obtained from Jackson Laboratory were used to breed SMA type I Smn^−/−;SMN2 mice for the in vitro experiments (Monani et al., 2000) as well as Smn^−/−;SMN2;SMNΔ7 (hereinafter named SMN∆7) for the in vivo experiments in postnatal pups (Le et al., 2005). Smn^+/+;SMN2/Smn^+/−;SMN2 or Smn^+/+;SMN2;SMNΔ7/Smn^+/−;SMN2;SMNΔ7 were used as controls. For control experiments, TrkB-deficient Ntrk2^−/− were bred from Ntrk2^+/− mice originally described elsewhere (Rohrer et al., 1999). Size of the litters varied between 4 and 10. Genotyping of knock-out and transgene alleles was performed using DNA obtained from individual mice tail biopsies. The following primers were used: Smn: 5′-CTG​GAA​TTC​AAT​ATG​CTA​GAC​TGG​CCT​G-3′ (forward), 5′-CAA​TCT​ATC​ACC​TGT​TTC​AAG​GGA​GTT​GTG​G-3′ (reverse 1), 5′-GAT​GTG​CTG​CAA​GGC​GAT​TAA​GTT​G-3′ (reverse 2); SMN2: 5′-CAT​ACC​TTA​AAG​GAA​GCC​AC-3′ (forward), 5′-AGG​TTC​TGA​GGT​CAG​AAC​AGC-3′ (reverse); SMNΔ7: 5′-TCC​ATT​TCC​TCC​TGG​ACC​AC-3′ (forward), 5′-ACCCATTCCACTTCC TTT-3′ (reverse); TrkB: 5′-ACT​GAC​ATC​CGT​AAG​CCA​GT-3′ (forward), 5′-ATG​TCG​CCC​TGG​CTG​AAG​TG-3′ (reverse 1), 5′- GGT​TCT​AAG​TAC​TGT​GGT​TTC​C-3′ (reverse 2).

The human PLS3 (hPLS3)-containing AAV9 overexpression construct was purchased from VectorBuilder (www.vectorbuilder.com). The open reading frame (ORF) was tagged with an HA-tag and inserted into a self-complementary (sc) AAV9 under a CMV promotor (Vector ID: VB201021-1415qev). As control, an EGFP-containing scAAV9 (Vector ID: VB191101-2319suw) was used. AAV9 gene delivery was performed at Carl-Ludwig-Institute for Physiology at Leipzig University as previously described (Simon et al., 2017). In brief, P0/P1 mice were anesthetized by isoflurane inhalation and injected in the right lateral ventricle of the brain with ∼5 × 10¹⁰ genome copies of AAV9 vectors in a PBS solution containing a vital dye (Fast Green; Sigma-Aldrich).

SMN∆7 mice and aged-matched littermate controls were sacrificed at P5 or P10-11, respectively, and the Transversus abdominis anterior (TVA) muscle was dissected from the abdominal wall in pre-warmed physiological solution (135 mM NaCl, 12 mM NaHCO₃, 5 mM KCl, 1 mM MgCl₂, 20 mM Glucose, 2 mM CaCl₂). For BDNF stimulation experiments, the muscle for BDNF treatment was placed in physiological solution containing 500 ng/ml BDNF and incubated for 30 min shaking on a heating plate, while the untreated muscle was kept in physiological solution without BDNF for 30 min. Subsequently, the medium was exchanged to physiological solution without CaCl₂ for 5–10 min prior to fixation of the muscle with cold methanol-free 4% formaldehyde (Pierce, Thermo Fisher Scientific) for 90 min. Following three washing steps with PBS, they were kept in PBS until processing with the immunohistochemistry.

All steps were performed on an orbital shaker at room temperature (RT), if not stated otherwise. In brief, muscles were incubated with 0.1 M glycine (in PBS) for 30 min, permeabilized with 1% Triton X-100 in (TBS-T) for 2 × 5 min, 2 × 10 min, and 2 × 30 min and incubated in 10% bovine serum albumin (BSA in TBS-T) for 3 h prior to antibody staining. Vitamin D-binding protein (DBP; Lee et al., 2013) was used for the visualization of G-actin. For this experiment, TVA muscles were first exposed to ice-cold acetone for 7 min and then incubated with 15 μg/ml DBP (345802; Merck, in TBS-T) for 2 h at RT. After postfixation with methanol-free 4% formaldehyde for 30 min, the muscles were washed thrice with TBS-T, permeabilized with 1% Triton X-100 in (TBS-T) for 2 × 30 min and incubated in 10% BSA (in TBS-T) for 3 h. Primary antibodies were incubated in 2.5% BSA and 0.5–1 % Triton X-100 (in TBS-T) for 3 d shaking at 4°C. Afterwards, samples were washed thrice with TBS-T and secondary antibodies were incubated in 2.5% BSA and 0.05% Triton X-100 (in TBS-T) for 1 h. After final washing steps with TBS-T for 90 min, muscles were mounted on glass microscope slides using Aqua-Poly/Mount (18606; Polysciences). The following primary and secondary antibodies were used: polyclonal rabbit Ca²⁺ channel P/Q-type alpha-1A antibody (Ca_v2.1; 1:250, 152 203, RRID:AB_2619841; Synaptic systems), polyclonal rabbit anti-DBP (1:150, PA5-19802, RRID:AB_10989680; Thermo Fisher Scientific), polyclonal chicken anti-Neurofilament H (1:1,000, AB5539, RRID:AB_11212161; Sigma-Aldrich), monoclonal rabbit anti-p-TrkB (1:250, 4621, RRID:AB_916186; Cell signaling), polyclonal guinea pig anti-Synaptophysin-1 (1:500, 101 004, RRID:AB_1210382; Synaptic systems), polyclonal goat anti-TrkB (1:300, AF1494, RRID:AB_2155264; R&D systems), Alexa Fluor 488 AffiniPure donkey anti-guinea pig IgG (H + L; 1:500, 706-545-148, RRID:AB_2340472; Jackson Immuno Research), Cy3 AffiniPure donkey anti-chicken IgY (IgG; H + L; 1:500, 703-165-155, RRID:AB_2340363; Jackson Immuno Research), Cy3 AffiniPure donkey anti-goat IgG (H + L; 1:500, 705-165-003, RRID:AB_2340411; Jackson Immuno Research), Cy3 AffiniPure donkey anti-guinea pig IgG (H + L; 1:500, 706-165-148, RRID:AB_2340460; Jackson Immuno Research), Cy3 AffiniPure donkey anti-rabbit IgG (H + L; 1:500, 711-165-152, RRID:AB_2307443; Jackson Immuno Research), Cy5 AffiniPure donkey anti-goat IgG (H + L; 1:500, Jackson Immuno Research, 705-175-003), Cy5 AffiniPure donkey anti-guinea pig IgG (H + L; 1.500, 706-175-148, RRID:AB_2340462; Jackson Immuno Research), Alexa Fluor 647 donkey anti-rabbit IgG (H + L) highly cross-adsorbed (1:600, A-31573, RRID:AB_2536183; Thermo Fisher Scientific). Postsynaptic nicotinic acetylcholine receptors were labeled with Alexa Fluor 488 α-Bungarotoxin (1:500, B13422; Invitrogen).

Spinal cord immunostainings were performed as previously described (Buettner et al., 2021). In detail, P10-P11 mice were perfused with PBS and 4% PFA followed by a 4% PFA postfixation over night at 4°C. On the next day, the spinal cord was briefly washed with PBS. Subsequently, the spinal lumbar L1 segment was dissected and embedded in warm 5% agar and serial transverse sections (75 μm) were cut at the vibratome. The sections were blocked with 5% normal donkey serum in 0.01 M PBS with 0.3% Triton X-100 (in PBS-T; pH 7.4) for 90 min and incubated with the primary antibodies over night at RT. On the next day, after six times of 10 min PBS washes, sections were incubated with the secondary antibodies diluted in PBS-T for 3 h. Following six washing steps with PBS for 10 min, sections were mounted on slides and cover-slipped with an anti-fading solution made of Glycerol:PBS (3:7). The following primary and secondary antibodies were used: Polyclonal goat anti-ChAT (1:500, AB144P, RRID:AB_2079751; Merck), polyclonal chicken anti-GFP (1:2,000, ab13970, RRID:AB_300798; Abcam), monoclonal rat anti-HA (1:500, 11867423001, RRID:AB_390918; Roche), Alexa Fluor 488 AffiniPure donkey anti-chicken IgY (IgG; H + L; 1:1,000, 703-545-155, RRID:AB_2340375; Jackson Immuno Research), Cy3 AffiniPure donkey anti-rat IgG (H + L; 1:1,000, 712-165-150, RRID:AB_2340666; Jackson Immuno Research), Alexa Fluor 647 AffiniPure donkey anti-goat IgG (1:1,000, 705-605-003, RRID:AB_2340436; Jackson Immuno Research).

To specifically knockdown endogenous Smn and Pls3 in motoneurons, shRNA oligonucleotides targeting the murine Smn and Pls3 were designed and introduced into a GFP-expressing pSIH-H1 shRNA vector (System Bioscience) according to the manufacturer’s protocol. As control, an empty pSIH-H1 shRNA vector (GFP) or Luciferase-containing pSIH-H1 shRNA vector (shLuci) was used. The following antisense sequences were used for cloning: Smn: 5′-GAA​GAA​TGC​CAC​AAC​TCC​C-3′; Pls3: 5′-CTC​ATG​GTG​GAT​GGT​GAC​AG-3′. For generation of the TrkB-GFP construct, the coding sequence of GFP was fused to the C-terminus of the murine TrkB coding sequence and inserted into a FuVal vector backbone under the ubiquitin promotor harboring a Kozak sequence. Correct plasmid insertion was checked by sequencing (LGC Genomics GmbH) and plasmid purification for lentivirus production was performed with the NucleoBond Xtra Midi plasmid purification kit (Machery Nagel). For overexpression of hPLS3, the constructs were obtained from VectorBuilder (www.vectorbuilder.com). The ORF was inserted under the ubiquitin promoter into a mammalian gene expression vector containing mCherry (Vector ID: VB200130-1089naj). As control, a mCherry containing vector (Vector ID: VB181226-1081ctz) was used. Lentivirus packaging was performed in HEK293TN cells transfected through TransIT-293 (Mirus) with the corresponding plasmid together with pCMV-VSVG and pCMVΔR8.91 helper plasmids (Rehberg et al., 2008). Upon ultracentrifugation at 25,000 rpm (SW 32 Ti rotor, Beckman) for 120 min at 4°C, the viral supernatant was harvested and stored at −80°C.

The primary motoneuron cell culture was performed according to Wiese et al. (2010). In detail, the lumbar part of the spinal cord was dissected from mouse embryos on embryonic day (E) 12.5 and digested with 0.1% trypsin (Worthington). After gentle trituration, the cells were enriched through anti-p75^NTR antibody panning (MLR2, M-009-100, RRID:AB_2492396; Biosensis) and plated onto poly-DL-ornithine hydrobromide (PORN; 0.5 mg/ml, P8638; Sigma-Aldrich) and laminin-221/211 (2.5 µg/ml, CC085; Merck) coated glass coverslips. For RNA and Western blot analysis, the cells were plated on PORN/laminin-111 (2.5 µg/ml, 23017-015; Thermo Fisher Scientific) coated 24-well plates. For live-cell imaging experiments, motoneurons were plated on PORN/laminin-221/211 coated 35 mm high µ-dishes (81156; Ibidi). Compartmentalized motoneuron cultures were maintained in microfluidic chambers (Xona Microfluidics, SND 150) pre-coated with PORN/laminin-111. For lentiviral transduction, the motoneuron suspension was incubated with the lentiviral particles for 15 min prior to plating. Motoneurons were maintained in complete medium (Neurobasal medium [NB]; Life Technologies) supplied with 500 µM Glutamax (Life Technologies), 2% B27 (Life Technologies), 2% heat-inactivated horse serum (Linaris), as well as neurotrophic factors CNTF (5 ng/ml) and BDNF (2 ng/ml). For compartmentalized chambers, a BDNF concentration gradient was established to achieve a directed axonal growth from no BDNF at the somatodendritic side to 20 ng/ml BDNF at the axonal side, while 5 ng/ml CNTF was added on both sides. The cells were kept at 37°C and 5% CO₂, and the medium was exchanged on the first day in vitro (DIV) and then every second or third day. Standard experiments were performed on DIV5. RT-PCR, Western blot analysis or experiments with lentiviral transduced cells were performed on DIV7.

For BDNF stimulation experiments, cells were grown in complete medium supplemented with BDNF (2 ng/ml) and CNTF (5 ng/ml). Depending on the experimental setup, on DIV4 or DIV6, motoneurons were deprived from BDNF for 16–24 h prior to the experiments. Cells were washed thrice with NB medium, and complete medium supplemented with CNTF (5 ng/ml) was added. On DIV5/DIV7, motoneurons were incubated with 100 ng/ml BDNF in complete medium for 15 min. Subsequently, cells were washed with NB medium and fixed with methanol-free 4% formaldehyde for 10 min.

Live-cell immunocytochemistry was performed to exclusively label TrkB presented at the plasma membrane. Motoneurons were cultured on laminin-221/211 coated glass coverslips and deprived from BDNF 1 d prior to the experiments as described before. On DIV5/DIV7, motoneurons were placed on parafilm and immediately covered with pre-warmed complete medium. According to the experimental design, cells were pre-treated with Cytochalasin D (C2618; 0.5 µM, 30 min; Sigma-Aldrich), Nocodazole (M1404; 10 µM; Sigma-Aldrich, 2 h), Cycloheximide (01810; 20 µM; Sigma-Aldrich, 1 h) or CK-666 (SML0006; 50 µM; Sigma-Aldrich, 30 min + 15 min during stimulation/recovery). For BDNF stimulation, the medium was exchanged with 100 ng/ml BDNF-containing medium for 5 min or 15 min at 37°C. For cAMP stimulation, the medium was exchanged with 100 µM 8-(4- Chlorophenylthio) adenosine-3′, 5′- cyclic monophosphate (8-CPT-cAMP) containing complete medium for 20 min at 37°C. Afterward, cells were washed with cold complete medium and placed on ice. The antibody against the extracellular domain of TrkB (1:200, polyclonal rabbit anti-TrkB, 07-225, RRID:AB_310445; Merck) was diluted in cold complete medium and added for 45 min. After washing with PBS, the secondary antibody (Alexa Fluor 488 AffiniPure donkey anti-rabbit IgG [H + L], 1:300, Jackson Immuno Research, 711-545-152, RRID:AB_2313584 or Alexa Fluor 647 donkey anti-rabbit IgG [H + L] highly cross-adsorbed, 1:400, A-31573, RRID:AB_2536183; Thermo Fisher Scientific) diluted in cold complete medium was incubated for 30 min. Following three washing steps with cold PBS cells were fixed with methanol-free 4% formaldehyde for 5 min and standard immunofluorescence protocol without permeabilizing agents like Triton X-100 was used to visualize intracellular proteins.

For double live-cell-labeling experiments, DIV5/DIV7 motoneurons, that were previously deprived from BDNF, were placed on parafilm and immediately covered with pre-warmed complete medium. The cells were washed with cold complete medium and placed on ice. TrkB antibody (1:200, polyclonal rabbit anti-TrkB, 07-225, RRID:AB_310445; Merck) diluted in cold complete medium was added for 30 min, followed by three washing steps with cold PBS and Alexa Fluor 488 AffiniPure donkey anti-rabbit IgG (H + L) incubation (1:300, 711-545-152, RRID:AB_2313584; Jackson Immuno Research) diluted in cold complete medium for 25 min. After a washing step with complete medium, motoneurons were removed from ice and 100 ng/ml BDNF-containing pre-warmed complete medium was added for 5 min at 37°C. After that, remaining antibodies were washed off twice with 1 mM EDTA in complete medium and cells were incubated with warm complete medium for 10 min for recovery. Subsequently, motoneurons were placed back on ice and incubated with the TrkB antibody (1:200, polyclonal rabbit anti-TrkB, 07-225, RRID:AB_310445; Merck) diluted in cold complete medium for 30 min. Again, following three washing steps with cold PBS, the cells were incubated with Alexa Fluor 647 donkey anti-rabbit IgG (H + L) highly cross-adsorbed (1:400, A-31573, RRID:AB_2536183; Thermo Fisher Scientific) diluted in cold complete medium for 25 min. Following three washing steps with cold PBS, cells were fixed with methanol-free 4% formaldehyde for 5 min and standard immunofluorescence protocol omitting permeabilization was used to visualize intracellular proteins.

Motoneurons were grown on laminin-221/211 coated glass coverslips until DIV5/DIV7, fixed for 5–10 min with methanol-free 4% formaldehyde and then washed thrice with PBS. Coverslips were placed on parafilm and covered with TBS-T before they were permeabilized with 0.3% Triton X-100 (except for TrkB stainings) for 20 min. For G-actin staining using DBP, motoneurons were exposed to ice-cold acetone for 5 min followed by incubation with 5 μg/ml DBP for 1 h at RT. Following three washing steps with TBS-T, they were incubated with 10% BSA in TBS-T for 1 h at RT. Primary antibodies were incubated in 1% BSA in TBS-T overnight at 4°C. The next day, cells were washed with TBS-T thrice, followed by incubation with the secondary antibodies in 1% BSA in TBS-T for 1 h at RT. Lastly, cells were washed thrice with TBS-T and mounted on glass microscope slides using Aqua-Poly/Mount (18606; Polysciences) for confocal imaging and ProLong Glass (P36980; Thermo Fisher Scientific/Invitrogen) for high-resolution imaging. The following primary and secondary antibodies were used: monoclonal mouse anti-Arp3 (1:200, A5979, RRID:AB_476749; Sigma-Aldrich), polyclonal guinea pig anti-Ca²⁺ channel N-type alpha-1B (Ca_vN; 1:500, 152 305, RRID:AB_2619845; Synaptic systems), polyclonal rabbit anti-DBP (1:500, PA5-19802, RRID:AB_10989680; Thermo Fisher Scientific), monoclonal rat anti-mCherry (1:1,000, M11217, RRID:AB_2536611; Thermo Fisher Scientific/Invitrogen), monoclonal mouse anti-myosin VI (1:250, GTX11095, RRID:AB_369989; GeneTex), monoclonal rabbit anti-p-Akt (1:500, 4060, RRID:AB_2315049; Cell signaling), polyclonal rabbit anti-Plastin 3 middle region (1:100, ARP56623_P050, RRID:AB_10642988; Aviva systems biology), monoclonal rabbit anti-p-TrkB (1:500, 4621, RRID:AB_916186; Cell signaling), polyclonal guinea pig anti-Synaptophysin-1 (1:1,000, 101 004, RRID:AB_1210382; Synaptic systems), polyclonal rabbit anti-TrkB (1:1,000, 07-225, RRID:AB_310445; Merck), Alexa Fluor 488 AffiniPure donkey anti-rabbit IgG (H + L; 1:600, 711-545-152, RRID:AB_2313584; Jackson Immuno Research), Alexa Fluor 488 AffiniPure donkey anti-guinea pig IgG (H + L; 1:500, 706-545-148, RRID:AB_2340472; Jackson Immuno Research), Cy3 AffiniPure donkey anti-guinea pig IgG (H + L; 1:500, 706-165-148, RRID:AB_2340460; Jackson Immuno Research), Cy3 AffiniPure donkey anti-mouse IgG (H + L; 1:500, 715-165-151, RRID:AB_2315777; Jackson Immuno Research), Cy3 AffiniPure donkey anti-rabbit IgG (H + L; 1:500, 711-165-152, RRID:AB_2307443; Jackson Immuno Research), Cy3 AffiniPure donkey anti-rat IgG (H + L; 1:500, Jackson Immuno Research, 712-165-150, RRID:AB_2340666), Cy5 AffiniPure donkey anti-guinea pig IgG (H + L; 1:500, 706-175-148, RRID:AB_2340462; Jackson Immuno Research), Cy5 AffiniPure goat anti-mouse IgG (H + L; 1:500, 115-175-146, RRID:AB_2338713; Jackson Immuno Research), Alexa Fluor 647 donkey anti-rabbit IgG (H + L) highly cross-adsorbed (1:800, A-31573, RRID:AB_2536183; Thermo Fisher Scientific). F-actin was visualized either with ActinGreen 488 ReadyProbes Reagent (1:50, R37110; Invitrogen), ActinRed 555 ReadyProbes Reagent (Rhodamine Phalloidin, 1:50, R37112; Invitrogen) or with Alexa Fluor 546 Phalloidin (1:50, A22283; Invitrogen).

Image acquisition of mounted samples was performed at RT with an Olympus Fluoview 1,000 confocal system equipped with a UPLSAPO 60×/1.35 NA oil objective, a FVD10 SPD spectral detector and the FV10-ASW 4.2 software. For motoneurons, single stack images were taken with 16-bit, 512 × 512-pixel resolution, and for NMJs 16-bit 0.5 µm z-stack images with 1,024 × 1,024-pixel resolution were acquired. Identical settings were applied within one set of experiments. Quantification was done with unprocessed images using Fiji and the “sum slices” projection was used for z-stacks. The mean gray value per area of the immunosignal was quantified in the region of interest (ROI) and the background signal was subtracted. For normalization, all values originating from one experiment were divided by the mean of the corresponding untreated control group to correct for inter-experimental variations. For TrkB recycling assay, the total number of TrkB-A488 and TrkB-A647 immunosignals was counted after linear brightness and contrast adjustments. TrkB-A647, as recovered receptors after BDNF stimulation were counted as the percentage of TrkB-A488, as endocytosed receptors that have been at the cell surface prior to BDNF stimulation. The amount of co-localizing recycled TrkB was determined with JaCoP plugin based on “Objects based methods” and “Geometrical centre” analysis and used to calculate the percentage of recycled TrkB (recovered receptors that previously have been on the cell surface). Illustration and statistical analysis were performed with GraphPad Prism 6. To obtain higher resolution pictures, structured illumination microscopy (SIM) was performed on a Zeiss ELYRA S.1 or Zeiss ELYRA 7 system, each equipped with a plan APO 63×/1.4 NA oil objective and sCMOS pco.edge cameras. Images were processed with the Zeiss software ZEN (ZEN 2012 SP5 FP3 for the ELYRA S.1 and ZEN 3.0 SR FP2 for the ELYRA 7 system, respectively) using automated standard settings. SIM images were corrected for chromatic aberration by channel alignment created with Tetraspecks (T14792; Thermo Fisher Scientific). “Maximum intensity” projected SI-images were generated with Fiji and Brightness/Contrast were adjusted for better visibility.

For analysis of actin dynamics, motoneurons were grown on PORN/laminin-221/211 coated 35 mm high µ-dishes (81156; Ibidi). On DIV7, cells were incubated with 100 nM SiR-actin (far-red and cell-permeable probe for F-actin visualization, tebu-bio, SC006) for 2 h. After washing with PBS, pre-warmed Tyrode’s solution (125 mM NaCl, 2 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, 30 mM glucose, 25 mM HEPES, pH 7.4) was added and chamber was placed into Tokai Hit stage incubator for imaging at 37°C with constant 5% CO₂ supply. Imaging was performed using the Nikon’s Eclipse TE2000 inverted epifluorescence microscope equipped with a plan APO VC 60×/1.4 NA objective, a perfect focus system and NIS-Elements AR 4.40.00 software. A fluorescent LED light with an excitation wavelength of 635 nm was used. 16-bit, 1,024 × 1,022-pixel images (2 × 2 binning) were acquired every 15 s over a time period of 20 min (exposure time 200 ms) with the ORCA Flash 4.0 V2 C11440-22C camera (Hamamatsu Photonics). For quantification of actin dynamics within single filopodia of growth cones Fiji was used to generate maximum intensity projections and identify the total length of each filopodia starting at the growth cone center. The ROI was transferred to the time-lapse video and a kymograph was generated using the corresponding multi kymograph plugin. The moved distance of single filopodia (y axis, length in µm) as well as the velocity (µm/min) of single filopodia movements (y axis, length in µm divided by x axis, time in minutes) derived from various growth cones were illustrated.

Calcium imaging was performed on DIV7 motoneurons plated on PORN/laminin-221/211 coated 35 mm high µ-dishes (81156; Ibidi). Cells were incubated with 5 µM Oregon Green 488 BAPTA-1, AM (O6807; Invitrogen, in Pluronic F-127; Thermo Fisher Scientific) or Cal-590 AM (ABD-20511; Biomol, in Pluronic F-127; Thermo Fisher Scientific) diluted in calcium imaging buffer (135 mM NaCl, 1 mM MgCl₂, 10 mM HEPES, 1 mM CaCl₂, 6 mM KCl, 5.5 mM Glucose, pH 7.4) for 15 min at 37°C. Upon three washing steps, cells were covered with calcium imaging buffer and placed into Tokai Hit stage incubator for imaging at 37°C with constant 5% CO₂ supply. Calcium imaging was performed at the Nikon’s Eclipse TE2000 inverted epifluorescence microscope equipped with a plan APO VC 60×/1.4 NA objective, a perfect focus system and NIS-Elements AR 4.40.00 software. A fluorescence LED light source was used for excitation at 470 nm (for Oregon Green) or 590 nm (for Cal-590 AM). 16-bit, 1,024 × 1,022-pixel pictures (2 × 2 binning) were taken at a frequency of 2 Hz over 5 min (exposure time 100 ms) at the ORCA Flash 4.0 V2 C11440-22C camera (Hamamatsu Photonics). The resulting 601 frame videos were analyzed with Fiji. Growth cones were defined as region of interest and a dynamic Z-axis profile, normalized against the average of the first 20 frames (F/F0), was plotted. Calcium spikes were counted using the BAR Plugin.

TrkB-GFP live-cell imaging was performed on motoneurons co-transduced with LV-TrkB-GFP and LV-mCh or LV-hPLS3 respectively, plated on PORN/laminin-221/211 coated 35 mm high µ-dishes (81156; Ibidi). ON DIV7, motoneurons were washed with pre-warmed Tyrode‘s solution (125 mM NaCl, 2 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, 30 mM glucose, 25 mM HEPES, pH 7.4) and incubated with 100 µM 8-(4- Chlorophenylthio) adenosine-3', 5'- cyclic monophosphate (8-CPT-cAMP) containing Tyrode‘s solution for recording. Imaging was performed at the Nikon’s Eclipse TE2000 inverted epifluorescence microscope equipped with a plan APO VC 60×/1.4 NA objective, a perfect focus system and NIS-Elements AR 4.40.00 software. A fluorescence LED light source was used for excitation at 470 nm. 16-bit, 1,024 × 1,022-pixel pictures (2 × 2 binning) were acquired every ∼ 2.5 s over a time period of 10 min (exposure time 100 ms) using the ORCA Flash 4.0 V2 C11440-22C camera (Hamamatsu Photonics). The resulting videos were analyzed with Fiji.

For RNA-seq, motoneurons were transduced with Smn knockdown or GFP control lentiviruses and cultured in microfluidic chambers (Xona Microfluidics, SND 150; Saal et al., 2014). On DIV7, RNA was extracted from the somatodendritic and the axonal compartments and whole-transcriptome profiling was used for cDNA amplification and library generation (Briese et al., 2016). The corresponding GFP control samples were described previously (Briese et al., 2018) and processed in parallel to the Smn knockdown samples described here. Libraries were sequenced on an Illumina MiSeq System and reads were mapped to the mouse genome with Star version 2.4.0d (https://code.google.com/p/rna-star/). Calculation of fragments per kilobase of transcript per million mapped reads (FPKM) values and analysis of differential gene expression was done using the Cufflinks package version 2.2.1 (http://cufflinks.cbcb.umd.edu/; Briese et al., 2016). For GO term analysis the Database for Annotation, Visualization and Integrated Discovery (DAVID) was used (Huang et al., 2009). As background datasets, the 10,433 and 11,127 transcripts previously detected in the somatodendritic and axonal compartment, respectively of wild-type motoneurons were used (Briese et al., 2016). The sequencing data from Smn knockdown motoneurons described in this publication are accessible in NCBI’s Gene Expression Omnibus through GEO Series accession no. GSE197638 (Edgar et al., 2002).

For Western blot analysis to quantify the total protein amount in the cells, motoneurons were cultured on PORN/laminin-111 coated 24-wells until DIV7. Cells were deprived from BDNF on DIV6 as previously described. On DIV7, cells were stimulated with 100 ng/ml BDNF for 15 min, washed with HBSS and directly lysed in Laemmli buffer (125 mM Tris, pH 6.8, 10% SDS, 50% glycerol, 25% β-mercaptoethanol and 0.2% bromophenol blue). Lysates were boiled for 5 min at 99°C, separated on a 4–12% gradient SDS-PAGE gel and transferred to a polyvinylidene fluoride membrane (BioRad). Unspecific binding sites were blocked with 5% milk (TBS-T) for 1 h and probed with the following primary antibodies in TBS-T over night at 4°C. Monoclonal rabbit anti-p-TrkB (1:1,000, 4621, RRID:AB_916186; Cell signaling), polyclonal rabbit anti-TrkB (1:1,000, 07-225, RRID:AB_310445; Merck), monoclonal mouse anti-beta actin (1:3,000, GTX26276, RRID:AB_367161; GeneTex). Following three washing steps, the membrane was incubated with horseradish-peroxidase (HRP)-conjugated secondary antibodies for 1 h at RT and immunodetection was performed using Amersham ECL Western blotting Detection Reagent (Cytiva) and x-ray films (Fuji super RX). The following secondary antibodies were used: peroxidase AffiniPure goat anti-rabbit IgG (H + L; 1:10,000, 111-035-144, RRID:AB_2307391; Jackson Immuno Research) and peroxidase AffiniPure goat anti-mouse IgG (H + L; 1:10,000, 115-035-003, RRID:AB_10015289; Jackson Immuno Research).

RNA from DIV7 motoneurons cultured on PORN/laminin-111-coated 24-wells was isolated using the NucleoSpin RNA kit (Macherey-Nagel) and transcribed in cDNA with the help of the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). RT-PCR was performed using 1:5 diluted cDNA and Luminaris HiGreen qPCR Master Mix (Thermo Fisher Scientific) on a LightCycler 1.5 thermal cycler (Roche). Absolute copy numbers of TrkB and Plastin 3 were normalized to the absolute copy numbers of Gapdh as reference gene. The following primers were used: murine TrkB: 5′-CGG​GAG​CAT​CTC​TCG​GTC​TAT-3′ (forward), 5′-CTG​GCA​GAG​TCA​TCG​TCG​TTG-3′ (reverse); murine Gapdh: 5′-GCA​AAT​TCA​ACG​GCA​CA-3′ (forward), 5′-GTC​GTG​GAG​TCT​ACT​GGT​G-3′ (reverse); murine Pls3: 5′-CAA​GCC​TCC​ATA​CCC​AAA​GC-3′ (forward), 5′-CCA​TCG​TTC​AGG​TCT​TGT​CC-3′ (reverse); human PLS3: 5′-GAA​CGT​TGA​GTG​AAG​CTG​GA-3′ (forward), 5′-TTG​CCA​CTC​TTC​ACA​AGG​TC-3′ (reverse).

For statistical analysis, GraphPad Prism 6 was used. For all data sets, at least three independent experiments were performed, and “N” represents the set of experiments while “n” indicates the number of independent data points. For n < 40, the data is shown as mean ± SD; for n > 40 the results are shown as mean ± SEM. Normal distribution was tested using D’Agostino & Pearson omnibus normality test. Two-tailed unpaired t tests were used for normally distributed data sets and group analyses (Smn/Pls3 FPKMs). For not equal distributed data sets, Mann–Whitney U-test was used for comparison of two groups while more groups were analyzed by one-way analysis of variance (ANOVA) Kruskal–Wallis test and Dunn’s Multiple Comparison post-hoc test. Significance is indicated as n.s. (not significant) if P > 0.05, * if P < 0.05, ** if P < 0.01, *** if P < 0.001, and **** if P < 0.0001. Data is shown as scatter dot plots with bars/bar plots for FPKMs.

The final processing of representative images and graphs was performed with Adobe Illustrator v 25.2.2 software.

Fig. S1 shows disturbed TrkB localization and phosphorylation upon BDNF stimulation in SMNΔ7 NMJs at P5 and Western blot analysis of TrkB and p-TrkB in control and SMA motoneuron whole-cell lysates. Fig. S2 demonstrates the antibody specificity of TrkB and p-TrkB antibodies used in this study. Fig. S3 shows that reduced profilin levels and p-profilin levels after BDNF stimulation in SMA motoneurons can be rescued by hPLS3 overexpression. Fig. S4 shows robust AAV9-delivered hPLS3 or GFP expression in the L1 spinal segment of injected P10 control mice. Table S1 contains the transcriptome analysis of the somatodendritic and axonal compartment of Smn knockdown and GFP-expressing control motoneurons. Table S2 contains the GO term analysis of significantly down- and upregulated transcripts in the somatodendritic and axonal compartment of Smn knockdown motoneurons. Videos 1, 2, and 3 show TrkB-GFP movements upon cAMP stimulation in Video 1 LV-mCh transduced control motoneurons, Video 2 LV-mCh transduced Smn-deficient motoneurons, and Video 3 LV-hPLS3 transduced Smn-deficient motoneurons.

We thank Nicole Rachor and Hildegard Troll for skilful technical support, Manas Kshirsagar for providing the TrkB-GFP lentiviral particles. The animal facility for the stock breeding of the mouse lines and Marcus Behringer for assistance with SIM.

Work in the authors’ lab is supported by the Deutsche Forschungsgemeinschaft, Grants JA 1823/3-1 to S. Jablonka, SE 697/6-1 to M. Sendtner, SI 1969/2-1 and SI 1969/3-1 to C.M. Simon, BR 4910/2–1 to M. Briese; by a grant from Picoquant to M. Sendtner; by Cure SMA (JAB1920) to S. Jablonka and SMA Europe to C. Simon.

Author contributions: S. Jablonka and M. Sendtner conceived the project and thereafter S. Jablonka, M. Sendtner and L. Hennlein designed the experiments. L. Hennlein performed most of the experiments including TVA preparations, clonings, motoneuron cultures, TVA and motoneuron immunostainings, image acquisition and live-cell imaging, Western blot analysis and RT-PCR analysis as well as analysis of the data. H. Ghanawi performed the cloning of shPls3, helped with motoneuron cultures, Western blot analysis and RT-PCRs. C.M. Simon and F. Gerstner performed the AAV9-treatment in neonatal mice, TVA preparations and spinal cord immunohistochemistry. E. Palominos-Garcia performed the motoneuron Ca_v2.2 immunostainings and calcium imaging and data analysis under the assistance of L. Hennlein. L. Saal-Bauernschubert performed the cloning of shSmn and the compartmentalized motoneuron cultures and their preparation for RNA-seq. S. Appenzeller and M. Briese performed bioinformatic analysis of the RNA-seq data. E. Yildirim helped with calcium imaging. M. Moradi and C. Deng provided the TrkB-GFP construct. M. Moradi helped with cloning and live-cell imaging. M. Sauer and T. Klein performed SIM on the ELYRA 7. L. Hennlein prepared the graphical abstract, figures, and additional files. L. Hennlein and S. Jablonka wrote the manuscript with the assistance of all co-authors.